|Abstract or Summary
- The illicit use of prescription stimulants (IUPS) is a critical Public Health problem in the college population that represents a unique form of substance use. Namely, the primary motives for IUPS by college students are academic in nature (e.g., Teter, McCabe, LaGrange, Cranford, & Boyd, 2006), which may explain why usage rates as high as 43% have been reported on college campuses (Advokat, Guidry, & Martino, 2008). As the field of IUPS is in its infancy, the IUPS-literature lacks 1) a consistent definition of "illicit use of prescription stimulants", 2) an instrument designed to assess the multiple influences on IUPS behavior, and 3) a theoretical lens (Bavarian, 2010a). Accordingly, the goal of this study was to address these research gaps by examining IUPS at one university located in the Pacific Northwest using the theory of triadic influence (TTI; Flay & Petraitis, 1994; Flay, Snyder, & Petraitis, 2009), an integrated, ecological approach to explaining and predicting health behaviors.
The study included five phases of research. The first phase began with a systematic approach to measurement development resulting in a 97-item preliminary instrument, the Behaviors, Expectancies, Attitudes and College Health Questionnaire (BEACH-Q). The measure was reviewed by the dissertation committee, leading to revisions in content and structure. During phase II, the BEACH-Q was evaluated by a convenience sample of five health and measurement professionals, and results suggested good content validity of the instrument (i.e. 35 of the 37 TTI-based covariates received a median rating between "agree" and "strongly agree", and 34 received a mean rating between "agree" and "strongly agree"). In phase III (instrument review by a convenience sample of six undergraduates), all 37 covariates received a median and mean rating between "agree" and "strongly agree", confirming good face validity. In phase IV, one undergraduate classroom was selected to participate in a pilot test of the BEACH-Q, using test-retest methodology. Results from the pilot (N = 39) showed that the instrument had moderate to high internal consistency reliability and modest to high stability reliability. In phase V, the final 96-item version of the BEACH-Q was administered throughout campus using one-stage cluster sampling, with classrooms as the sampling unit and students as the observation unit (N = 520 students in 20 classrooms, eligible student response rate = 96.30%).
The prevalence of IUPS during college self-reported by the phase V sample was 25.58%, and the probability of engaging in IUPS did not differ between classrooms (Median Odds Ratio = 1.00). In crude logistic regressions, 24 out of 37 hypothesized theoretical covariates were significantly associated with IUPS in the expected direction, suggesting predictive validity of the BEACH-Q. Nested logistic regression analyses illustrated that, for the full multivariate model including ultimate, distal, and proximal covariates, the following measures in the intrapersonal stream were significantly associated with IUPS during college: race/ethnicity, year in school, academic concern, academic grades, diagnosis with Attention Deficit Disorder, and IUPS avoidance self-efficacy. Significant social situation/context stream covariates in the full model included: residence, varsity sports participation, perceptions of IUPS by friends, family, and faculty, endorsement of IUPS by friends, and perceived prevalence of IUPS among friends. With
respect to the sociocultural environment stream, in the full multivariate model, the following covariates were found to be significantly associated with IUPS: financial-related stress, participation in religious activities, positive IUPS expectancies, prescription stimulant knowledge and perceived costs/benefits of IUPS. Lastly, intention to engage in IUPS (an immediate precursor) was positively associated with IUPS.
Structural equation modeling was used to test models of IUPS for each of the three streams, as well as one integrated model that included covariates from each stream. The models all demonstrated good model fit, and provided insight into the factors that influence (and suggest the mechanisms of causation) intentions to engage in, and ultimately the behavior of, IUPS. In the intrapersonal stream model, inattention was positively associated with academic concern and inversely associated with avoidance self-efficacy, and avoidance self-efficacy was inversely associated with intentions to engage in IUPS. Moreover, intentions to engage in IUPS and avoidance self-efficacy were both significantly associated with IUPS. In the social situation/context stream model, living on-campus was negatively associated with friends' endorsement of IUPS which was positively associated with perceived prevalence of IUPS among friends, and perceived prevalence was positively associated with intentions to engage in IUPS; all of the direct paths to IUPS, excepting the path from perceived prevalence of IUPS among friends, were significant. In the sociocultural environment model, perception of course demand was significantly associated with both the perception that professors give the most attention to top academic students and attitudes about the impact of prescription stimulants on academics, and these attitudes had a positive association with intentions to engage in IUPS. Moreover, the direct paths from attitudes and intentions to IUPS were significant. The mixed model, including participation in religious activities (sociocultural environment stream), friends' endorsement of IUPS (social situation/context stream), and avoidance self-efficacy (intrapersonal stream), also had significant paths from ultimate to distal to proximal to immediate precursor, and significant direct paths to IUPS.
This study successfully achieved its goals. First, the instrument developed was theory-driven, broadly defined IUPS, and was psychometrically strong. The cross-sectional study illustrated that IUPS was prevalent on the campus under investigation, as one in four students had engaged in the behavior during college. Associations of use with theoretical correlates were tested for, and structural equation modeling was used to support one premise of the TTI (i.e. that behavior is multifaceted, and covariates from different streams may interact to influence behavior). The findings also suggest that prevention and intervention plans should be multifaceted in nature. Given that this study's cross-sectional nature limited the ability to make causal inferences, future research involving the BEACH-Q should use longitudinal designs.